Plant species classification using deep英文文献资料.pdf

b i o s y s t e m s e n g i n e e r i n g 1 5 1 ( 2 0 1 6 ) 7 2 e8 0 Available online at ScienceDirect journal homepage: /l ocate/issn Research Paper Plant species classification using deep convolutional neural network Mads Dyrmann a, *, Henrik Karstoft b, Henrik Skov Midtiby a a The Maersk Mc-Kinney Moller Institute, University of Southern Denmark, Denmark b Department of Engineering, Aarhus University, Denmark a r t i c l e i n f o Information on which weed species are present within agricultural fields is important for Article history: site specific weed management. This paper presents a method that is capable of recog- Received 16 March 2016 nising plant species in colour images by using a convolutional neural network. The Received in revised form network is built from scratch trained and tested on a total of 10,413 images containing 22 22 August 2016 weed and crop species at early growth stages. These images originate from six different Accepted 26 August 2016 data sets, which have variations with respect to lighting, resolution, and soil type. This includes images taken under controlled conditions with regard to camera stabilisation and illumination, and images shot with hand-held mobile phones in fields with changing Keywords: lighting conditions and different soil types. For these 22 species, the network is able to Plant classification